
原文
GPT Image 2 · en
Scene:
Create a square 1:1 portrait image for Relationship Portrait Series. Use a natural-light or soft practical-light frame with believable candid energy and controlled composition. The frame should feel like a finished professional deliverable rather than a casual snapshot, with clear lens discipline, believable light falloff, and a palette that directly supports Photograph pairs or small groups in ways that reveal intimacy, tension, friendship, or shared history through pose, distance, and gaze.. Keep enough surrounding space and compositional control for gallery, research, campaign, or approval-board reuse while staying photoreal and human-scaled.
Subject:
Center the same recurring pair: a woman with golden-brown skin and a curly bob, and her partner with warm brown skin, cropped hair, and a relaxed upright build. Keep the same pair exact while opening the mood into laughter or a lighter candid beat. Use the example direction from the brief as grounding context: Photograph pairs or small groups in ways that reveal intimacy, tension, friendship, or shared history through pose, distance, and gaze.. Wardrobe, lighting, environment, and pose may shift to serve this variation, but the viewer must immediately understand that the same reference identity is still in front of the camera.
Important details:
This backlog line is difficult because Photograph pairs or small groups in ways that reveal intimacy, tension, friendship, or shared history through pose, distance, and gaze.. Preserve skin tone family, bone structure, eye spacing, nose shape, mouth shape, hairline logic, shoulder width, and overall body read across the entire series. preserve exact face / hair base / body proportions from the reference model. Even candid emotion must not loosen identity consistency. Avoid the common failure described in the brief: ordinary image models often lose likeness once styling, lighting, or layout changes enter the frame. Success should feel like the result should feel professionally usable, coherent, and clearly built around one stable human identity.
Use case:
Relationship Portrait Series series study 5/9, a candid shared-laugh portrait.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, preserve exact face / hair base / body proportions from the reference model, no random bystanders, no exaggerated open-mouth distortion, no new wardrobe family, no face drift.使い方
- パラメータを含む元のプロンプト全体をコピーします。
- 被写体、文字、構図を自分の要件に置き換えます。
- 対応する外部ツールで使用し、利用条件と結果を確認します。
開始前の確認
- 参考画像の要否を確認してください。互換モデルの表示は、そのモデルでの生成実績を意味しません。
- 名前、数値、画像内の文字を確認します。モデルの版で結果は変わります。





